deploy(mcp): MCP-server README (docker frontmatter) over updated trunk

Trunk now includes the table-import performance fix (scoped TS connection +
monthly-rotating demo database): table creates drop from ~250s/call to <1s,
so MCP builds shed ~8-10 minutes. Only README differs from develop_dg.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>

README.md

@@ -1,175 +1,41 @@
 ---
-title: ThoughtSpot Demo Builder
-emoji: 🚀
+title: DemoPrep MCP
+emoji: 🔌
 colorFrom: blue
-colorTo: purple
-sdk: gradio
-sdk_version: "4.44.1"
-app_file: app.py
+colorTo: green
+sdk: docker
+app_port: 7860
 pinned: false
 license: mit
-python_version: "3.11"
 ---
 
-# DemoPrep — AI-Powered ThoughtSpot Demo Builder
+# DemoPrep MCP Server
 
-A Gradio application that generates and deploys complete ThoughtSpot demo environments end to end — company research, an LLM-authored data blueprint, deterministic data generation, a Snowflake schema + data load, and a ThoughtSpot semantic model and liveboard.
+An MCP (Model Context Protocol) server that exposes DemoPrep's demo-build pipeline
+as tools an external agent (ThoughtSpot **AgentSpot**) can call — building a full
+demo (Snowflake schema → ThoughtSpot model → liveboard) from a prospect **brief**,
+with no human clicking through the Gradio UI.
 
-> **Deployed on Hugging Face Spaces.** `app.py` is the entry point. See [DEPLOYMENT.md](DEPLOYMENT.md) for setup.
+- **Transport:** streamable-HTTP at `/mcp`, gated by `Authorization: Bearer <MCP_ACCESS_TOKEN>`.
+- **Health:** unauthenticated `GET /` returns `{"status":"ok"}`.
+- **Tools:** `ping` (connectivity check) and `build_demo_from_brief` (full build).
 
-## 🖥️ The interface
+This Space shares the DemoPrep codebase with the app Space but runs the MCP
+entrypoint (`mcp_server.py`) instead of the Gradio app — the business logic lives
+in one place (`demoprep_app/` + the controller). Builds run for minutes, so the
+tool is meant to be driven job-style (fire, then check back).
 
-The app is a **tabbed Gradio UI**, and the **App tab is the primary way to build a demo**: fill in a short form and press **GO**, and the full pipeline runs automatically with live progress.
+## Required secrets
 
-> **Naming note:** the whole application lives in **`chat_interface.py`** — that name is *historical*. It is **not** just a chat interface; that single file hosts every tab (App, Chat, Admin, Settings, Run History, …), and the **App tab is the main path**. A **Chat tab** offers an alternative conversational flow that runs the same pipeline. `app.py` simply configures and launches this app.
+Set these in **Settings → Repository Secrets** (nothing sensitive is committed):
 
-## 🚀 Features
+| Secret | Purpose |
+|--------|---------|
+| `SUPABASE_URL`, `SUPABASE_ANON_KEY` | bootstrap → pulls Snowflake + LLM keys from admin settings |
+| `OPENAI_API_KEY` | LLM key (env-only) |
+| `TS_ENV_1_LABEL`, `TS_ENV_1_URL`, `TS_ENV_1_KEY_VAR` | the target ThoughtSpot environment |
+| `MCP_OWNER_EMAIL` | TS user that owns created objects (trusted-auth) |
+| `MCP_TS_ENV_LABEL` | which TS env label to deploy into |
+| `MCP_ACCESS_TOKEN` | shared bearer secret the calling agent presents |
 
-- **AI-powered research** — researches the company and industry with an LLM
-- **Single blueprint pipeline** — one path, no keyword routing or fallback: research → LLM-authored `DemoBlueprint` → deterministic engine → validation → derived DDL → load
-- **Realistic data** — bounded/related measures (a part never exceeds its whole, rates stay ≤ 1, prices are stable per entity) plus planted, discoverable "Demo to Win" insights the validator proves are visible
-- **ThoughtSpot integration** — deploys the connection, tables, semantic model, and an enhanced liveboard
-- **Interactive UI** — tabbed Gradio app with real-time progress
-
-## 🛠️ Tech Stack
-
-- **Frontend**: Gradio (Python web UI)
-- **Backend**: Python 3.11
-- **Database**: Snowflake (keypair auth)
-- **Analytics**: ThoughtSpot (per-environment trusted auth)
-- **AI**: multiple LLM providers (OpenAI and Anthropic/Claude), routed via `llm_config.py`
-- **Settings**: Supabase (admin + per-user settings)
-
-## 📋 Prerequisites
-
-- Python 3.11
-- Snowflake account (keypair configured)
-- ThoughtSpot Cloud account
-- An LLM API key (OpenAI and/or Anthropic)
-- Supabase project (settings storage)
-
-## 🚀 Quick Start
-
-1. **Clone the repository**
-   ```bash
-   git clone <repo-url>
-   cd demoprep
-   ```
-
-2. **Set up a virtual environment**
-   ```bash
-   python -m venv .venv
-   source .venv/bin/activate   # Windows: .venv\Scripts\activate
-   ```
-
-3. **Install dependencies**
-   ```bash
-   pip install -r requirements.txt
-   ```
-
-4. **Configure bootstrap environment variables** (see Configuration)
-   ```bash
-   cp .env.example .env
-   # edit .env
-   ```
-
-5. **Run the application**
-   ```bash
-   python app.py
-   ```
-   `app.py` launches the full Gradio app defined in `chat_interface.py`.
-
-6. **Open your browser**
-   Navigate to `http://localhost:7860` and use the **App** tab.
-
-## ⚙️ Configuration
-
-`.env` holds only the **bootstrap** secrets needed to start the app and reach Supabase and the LLM providers. All other credentials — the Snowflake keypair and the ThoughtSpot **per-environment trusted-auth keys** — live in **Supabase admin settings** and are loaded at runtime.
-
-```env
-# Supabase (settings storage)
-SUPABASE_URL=...
-SUPABASE_ANON_KEY=...
-
-# LLM providers
-OPENAI_API_KEY=...
-GOOGLE_API_KEY=...            # optional (Gemini)
-
-# Slack deployment notifications (optional, outbound-only)
-SLACK_BOT_TOKEN=xoxb-...
-SLACK_DEPLOYMENT_CHANNEL_ID=C0123456789
-```
-
-Slack notifications use the Slack Web API to post deployment status into one approved channel. This path is outbound-only: no Socket Mode, event subscriptions, slash commands, or public request URL. The Slack app needs only the `chat:write` bot scope and must be invited to the target channel.
-
-## 🎯 Usage
-
-The **App tab** is the primary interface — fill the form, press **GO**, and the pipeline runs end to end:
-
-1. **Fill the form** — vertical / line of business / function (or a custom use case), the company URL, and the target **ThoughtSpot environment**
-2. **Press GO** — research → blueprint → data generation → validation → Snowflake load → ThoughtSpot model + liveboard
-3. **Review** — model and liveboard links, a Demo Pack, and a Spotter Viz story appear on completion
-
-## 📁 Project Structure
-
-```
-demoprep/
-├── app.py                    # Entry point — launches the Gradio app on :7860
-├── chat_interface.py         # The full Gradio app (App/Chat/Admin/Settings/…) — historical name
-├── thoughtspot_deployer.py   # ThoughtSpot deploy: connection, tables, model, liveboard
-├── liveboard_creator.py      # MCP liveboard creation + TML post-processing (enhance_mcp_liveboard)
-├── snowflake_auth.py         # Snowflake keypair authentication
-├── llm_config.py             # LLM provider/model routing (single source of truth)
-├── demo_personas.py          # Vertical × Function use-case configs
-├── demoprep_app/             # The demo-generation pipeline package
-│   ├── pipeline/build_demo.py    #   the single build entry point
-│   ├── scenario/                 #   blueprint contract + LLM authoring + directives
-│   ├── dataset/                  #   deterministic engine + validator
-│   ├── ddl/                      #   Snowflake DDL derived from the dataset
-│   └── integrations/snowflake/   #   Snowflake row loader
-├── requirements.txt          # Python dependencies
-├── docs/                     # Documentation
-├── tests/                    # Unit tests + the e2e quality harness (e2e_quality.py)
-└── results/                  # Generated demo results
-```
-
-## 🧪 Testing
-
-```bash
-# Unit tests (pytest collects test_*.py)
-python -m pytest tests/
-
-# End-to-end quality harness — drives a running app via the browser and grades the output
-python tests/e2e_quality.py --env-name test
-```
-
-## 🤝 Contributing
-
-1. Fork the repository
-2. Create a feature branch (`git checkout -b feature/amazing-feature`)
-3. Commit your changes (`git commit -m 'Add amazing feature'`)
-4. Push to the branch (`git push origin feature/amazing-feature`)
-5. Open a Pull Request
-
-## 📄 License
-
-This project is licensed under the MIT License - see the [LICENSE](LICENSE) file for details.
-
-## 🙏 Acknowledgments
-
-- ThoughtSpot for the analytics platform
-- Snowflake for the data warehouse
-- OpenAI and Anthropic for the AI capabilities
-- Gradio for the web interface
-
-## 📚 Development Notes
-
-Sprint planning and working notes live in `dev_notes/` (gitignored); architecture and handoff docs are in `docs/`.
-
-## 📞 Support
-
-For support, create an issue in this repository.
-
----
-
-**Built with ❤️ for the ThoughtSpot community**
+`MCP_TRANSPORT=http` and `MCP_HTTP_PORT=7860` are baked into the Dockerfile.